Jiawei Liu

@stubbb Yeah, but I think in @salvador-blanco 's case, it might not be the issue, as he got segmentation fault even if using a batch size of 1. https://github.com/tensorlayer/hyperpose/issues/323#issuecomment-704679470

@salvador-blanco Hi, it seems like there are little useful details in your logging. So could you please just re-compile this program using "debug" flags, say '-Og' that it ensures very...

@Gyx-One Could you look at the Python part?

Hi @korejan , thanks for your advice. I will enhance the documents according to your feedbacks. When doing batching, we'll convert the images bytes into float type. And we did...

@AmrutaAnalytics Then you need to compile the C++ source files for it. In our project, we use python for development and c++ for deployment.

Hi @jinfagang , I am confused by your question. Did you mean that have we enabled tensorflow2.0 as the DNN backend? Currently, we only have TensorRT as the DNN backend,...

We don't have official guidance to build it on windows. Though I knew that some users have built it on Windows since all of our codes are cross-platform. I would...

> > We don't have official guidance to build it on windows. Though I knew that some users have built it on Windows since all of our codes are cross-platform....

we do 18 coco keypoints. Notably, when supporting openpifpaf, we convert its format to approximate 18 coco format. As for extending a Json output, you can modify the source codes...

@zhanghongyong123456 See https://github.com/tensorlayer/hyperpose/blob/master/src/stream.cpp#L133 I would appreciate it if you can contribute a JSON writer.